Self-Supervised Learning of Audio Representations using Angular Contrastive Loss

Keyphrases

• Contrastive Loss 100%
• Audio Representation 100%
• Self-supervised Learning 100%
• Learned Features 33%
• New Loss Function 33%
• Feature Embedding 33%
• Performance Improvement 16%
• Supervised Learning 16%
• Unsupervised Learning 16%
• Discriminative Ability 16%
• Linear Combination 16%
• Feature Representation 16%
• Sound Event Classification 16%
• Contrastive Learning 16%
• Downstream Task 16%
• Similar Samples 16%
• Instance Discrimination 16%
• Pretext Task 16%
• Angular Margin 16%
• Inherent Defects 16%
• Angular Margin Loss 16%

Computer Science

• Contrastive Loss 100%
• Self-Supervised Learning 100%
• Supervised Learning 16%
• Experimental Result 16%
• Unsupervised Learning 16%
• Linear Combination 16%
• Representation Learning 16%
• Learned Feature 16%
• Contrastive Learning 16%